Microsoft Foundry IQ is a knowledge layer for AI agents that creates reusable, multi-source knowledge bases, uses language-model-driven agentic retrieval to issue parallel subqueries and aggregate results, and returns permission-trimmed, traceable, grounded outputs.
Published May 16, 2026
| Starting Price | Usage-based (Azure AI Search capacity + agentic retrieval tokens + model tokens); no single public base monthly price for Foundry IQ alone | Managed enterprise search with an hourly-index pricing model | Vector database pricing with published minimums for certain plans |
| Most appropriate for | Microsoft-centric enterprises needing permission-aware, multi-source retrieval integrated with Entra/Purview | AWS-centric organizations seeking a managed enterprise search service with an hourly-index pricing model | Teams needing a dedicated vector database and control over the rest of the RAG/agent pipeline |
| Key Strength | Multi-source knowledge bases with ACL-aware retrieval and agentic retrieval planning/synthesis within a Microsoft technology stack | Managed index model and search capabilities tailored for AWS environments | Specialization in vector search infrastructure, with clear plan tiers and enterprise controls |
| Key Weakness | Preview status, pricing complexity, dependence on multiple Azure components, and ecosystem naming complexity | Costs can grow with continuously running indexes and feature fit depends on document volume and query load | Not a standalone enterprise knowledge layer; connectors, ACL trimming, governance, and agent orchestration typically require additional implementation |
—